some thoughts on panama/jextract

[Maurizio Cimadamore]

Thanks Marco,
one of the issues here is that we're effectively in the process of 
rebasing the foreign support on more lower-level building bricks - this 
means that, at least for some weeks, there will be confusion about the 
old vs. new API, and what 'jextract' exactly means. I hope we will be 
able to push the first version of the new, minimal jextract as soon as 
possible, so that we can move forward with 'deprecating' the 'foreign' 
branch - which in itself I think will remove a lot of confusion. The 
second step, as I said, will be to ship binary snapshot with the new 
approach, so that people can try it out for good. It doesn't help that 
the discussion on this mailing list are referring to things that aren't 
even available in the public EA binaries.

But this is what happens when discussing changes that are in flux - and 
when we try to share things sooner rather than later; in a way, we're 
doomed either way - if we share soon, there's an increased risk of 
confusing developers - if we share late, we avoid some of the confusion, 
but we create more frustration, as people cannot see where we're headed.

I'll add this to my "lesson learned" diary :-) with a view to try and do 
best next time around.

Thanks
Maurizio

On 12/01/2020 14:58, Marco Hutter wrote:
> But as you said in a later mail: It's hard to...
> 1. initially get a grip on such a large project and
> 2. keep track of the discussions and related changes that are done by 
> the maintainers
> For example, I did these experiments also before the major revamping 
> that Maurizio summarized in 
> http://cr.openjdk.java.net/~mcimadamore/panama/jextract_distilled.html 
> - I still have to catch up with that, and give it another try with the 
> latest EA build.